More comments below. Cheers, Casey
1. The definitions of iterative mode and recursive mode still seem a bit awkward. Also, the bit about answering from cache doesn't seem to be a part of recursive mode, but of general response to queries. And the discussion of the distinction between recursive servers and recursive servers is not useful, nor do I think it is accurate. I suggest the following: Iterative mode: A resolution mode in which, when a client queries a server, "the server refers the client to another server and lets the client pursue the query" (RFC 1034, section 2.3). Also called "non-recursive mode" (RFC 1034, section 4.3.1). A resolver (i.e., the "client") that works in iterative mode is sometimes called an "iterative resolver". Recursive mode: A resolution mode in which, when a client queries a server, "the first server pursues the query for the client at another server" (RFC 1034, section 2.3). A resolver (i.e., the "server" queried by the "client") operating in this mode is commonly called a "recursive server" or "recursive resolver". RFC 1123 also refers to such as a "recursive name server". 2. Some re-wording of full-service resolver for clarity: Full-service resolver: "A complete implementation of the resolver service", capable of iterative-mode resolution, configured with root server information, and implementing a cache, among other robustness requirements (RFC 1123, section 6.1.3.1). 3. More "quoted from" text that should be replaced with the appropriate use of quotation marks and parenthetical references: - Negative caching - Authoritative server - Authoritative-only server - Primary master. - Stealth server - Hidden master 4. I propose that the following paragraph be added to the definition of authoritative servers (alternatively they could be added as new terms): The terms "name server" and "domain name server" are both used to refer to authoritative servers (RFC 1034 section 2.4; RFC 1123, section 6.1.1). Occasionally the term "name server" is used more generically to refer to a server server responding to either recursive or non-recursive queries (RFC 1034, section 4.3.2). 5. Definition of authoritative-only server: s/ignore/ignore[s]/ s/It will not/[It] will not/ 6. In the definition of primary master, it seems more intuitive to switch the order of the paragraphs in the definition. Plus there is some redundancy in the definition (MNAME sentence used twice). And there should be a reference to "primary": "Master [or primary] server at the root of the AXFR/IXFR dependency graph. The primary master is named in the zone's SOA MNAME field and optionally by an NS RR. There is by definition only one primary master server per zone" (RFC 2136, section 1). The idea of a primary master is only used by RFC 2136, and is considered archaic in other parts of the DNS. 7. Definition of stealth server: s/slave/[secondary]/ 8. In the definition of forwarding, I think it is more accurate to distinguish the servers as "recursive resolvers", so it's clear we're not talking about stubs (this eliminates the need for the bit about not blindly relaying queries, in my opinion). See also the changes to forwarder below. Here's a proposed re-wording of the first paragraph: The process of one recursive resolver sending a DNS query with the RD bit set to another recursive resolver to resolve the query. A recursive resolver might be capable of forwarding to another recursive resolver, performing iterative resolution itself, or a combination or these, possibly on a per-name basis. [Second paragraph unchanged] 9. In the definition of forwarder, it seems intuitive to change the order of the sentences. Also, it seems to me that the discussion at the end of the text really has to do with forwarding referring to one of two different roles in the forwarding process. Forwarder: "A nameserver used to resolve queries instead of directly using the authoritative nameserver chain. The forwarder typically either has better access to the internet, or maintains a bigger cache which may be shared amongst many resolvers" (RFC 2308, section 1). This definition indicates that the forwarder is the recursive resolver to which queries are forwarded by another recursive resolver (see "Forwarding"). It is common today, however, for forwarder to refer to the recursive resolver forwarding the queries to the upstream recursive resolver. 10. On the definition of open resolver: s/processes queries/processes recursive queries/ s/stub resolver/resolver/
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop